Dram Oxide Screening Methods
Abstract
The objective of this program was to evaluate 'non-aging' screens that characterize maximum refresh times versus supply voltage in DRAMs. Extensive refresh characterizations were done, followed by accelerated life testing of DRAMs with periodic refresh characterization. The characterizations defined 1000 worst-case and 1000 random memory cells per part as the freak and main population of refresh time and measured the sensitivity of their refresh times to voltage. Accelerated life testing at high voltage stress was performed in an attempt to generate oxide-related failures. Six parts from two vendors were tested for individual bit variable hold time (VHT). Although we did not observe any TDDB failures during the life tests, our test results demonstrate that the 64K DRAM technology is highly reliable. Since only one recoverable failure (Vendor 2 S/N 338) occurred after 757,000 part hours (316,000 hours for vendor 1 and 441,000 hours for vendor 2), at near catastrophic voltages, we were unable to correlate initial bit refresh time with lifetime.
Document Details
- Document Type
- Technical Report
- Publication Date
- Dec 01, 1990
- Accession Number
- ADA231803
Entities
People
- Douglas R. Younkin